1. Technical Field of the Invention
The present invention generally relates to silyl functional polymers, to compositions containing same and to processes for the production thereof. More specifically, this invention relates to silyl substituted polycycloolefins that have been addition polymerized from polycycloolefin monomers containing at least one norbornene moiety. The polymeric compositions of this invention are useful in molded articles, in films, casted as articles, in coatings, as adhesives, and in membranes.
2. Background Art
Inorganic materials such as silicon dioxide and silicon nitride have been traditionally used in the microelectronics industry as insulating and passivating materials in the manufacture of integrated circuits. However, as the demand for smaller, faster, and more powerful devices becomes prevalent new materials will be needed to enhance the performance and the efficient manufacture of these devices.
To meet these enhanced performance and manufacturing criteria considerable interest in high performance polymers characterized by low dielectric constant, low moisture uptake, good substrate adhesion, chemical resistance, high glass transition temperatures (e.g., T.sub.g &gt;250.degree. C.), toughness, high thermo and thermo-oxidative stabilities, as well as good optical properties is increasingly gaining momentum. Such polymers are useful as dielectric coatings and films in the construction and manufacture of multichip modules (MCMs) and in integrated circuits (IC), in electronic packaging, in flexible film substrates, and in optical applications such as in flat panel displays and the like. Presently, substantial attention is being directed to polyimides and bis-benzocylobutenes (BCBs) for use as dielectric materials in the construction and manufacture of microelectronic devices.
Polyimides have been used as microelectronic dielectric materials because of their thermal and oxidative stability, high glass transition temperatures, and generally good mechanical properties. However, these materials have drawbacks which limit their performance, namely, (1) they often show considerable water uptake resulting in conductor corrosion and increased dielectric constant; (2) the electrical properties can be strongly anisotropic in fully cured and densified films, i.e., in-plane dielectric constant differs substantially (up to 50%) from through-plane dielectric constant; (3) the polyimide precursor, polyamic acid, requires the use of passivation layers to prevent unwanted reactions with copper conductors and substrates; (4) the dielectric constants are higher than desired; and (5) poor adhesion to copper and noble metal (i.e., gold, silver, and platinum) conductors. Moreover, polyimides must be cured at elevated temperatures (around 300.degree. C.) to effect ring closure of the polymer.
Benzocyclobutenes are also commercially employed in microelectronic packaging as insulating polymers. BCB's offer lower moisture absorption and lower dielectric constant values than the polyimides. However, they suffer the same adhesion drawbacks as the polyimides. Interfacial strength is quite low, consequently, films peel off copper and noble metal substrates with minimal effort. BCBs must also be cured at elevated temperatures in order to obtain polymers with useful physical properties. The physical properties of the cured polymer are dependent upon the cure time and temperature, i.e., the cure time and temperature determines the amount of crosslinking. In particular, BCBs gain their T.sub.g properties from crosslinking and hence suffer problems endemic to most thermosets in that vitrification limits the glass transition to approximately 20.degree. C. below the cure temperature due to reactant mobility restrictions. The physical properties of these polymers are therefore very dependent on the cure profile employed, necessitating microelectronic device manufacturers to precisely control and monitor device manufacture to obtain consistent polymer properties.
Polycycloolefins (e.g. polymers derived from polycyclic monomers containing a norbornene moiety) are well known in the art. Because of their high hydrocarbon content, polycycloolefins have low dielectric constants and a low affinity for moisture. Presently, there are several routes to polymerize cyclic olefin monomers such as norbornene or other higher polycyclic monomers containing the norbornene moiety. These include: (1) ring-opening metathesis polymerization (ROMP); (2) ROMP followed by hydrogenation; (3) addition copolymerization (Ziegler type copolymers with ethylene), and (4) addition homopolymerization. Each of the foregoing routes produces polymers with specific structures as shown in the Diagram 1 below: ##STR4##
As illustrated in the foregoing diagram a ROMP catalyzed polymer contains a repeat unit with one less cyclic unit than did the starting monomer. The so-called ring-opened repeat units are linked together in an unsaturated backbone characteristic of a ROMP polymer. As can readily be surmised ROMP catalyzed polymers suffer the inherent disadvantage of backbone unsaturation which significantly reduces the thermo-oxidative stability of the polymer. ROMP catalyzed polymers exist as thermoplastics or thermosets (T.sub.g &lt;240.degree. C.). ROMP catalyzed thermosets have been utilized to produce circuit board substrates via reaction injection molding (RIM) as disclosed in U.S. Pat. No. 5,011,730 to Tenney et al. However, as discussed above, these polymers inherently suffer from thermo-oxidative instability as well as insufficiently low T.sub.g s. Moreover, in the RIM process a finished polymer part is directly polymerized in the mold from a reactive monomer solution containing a molybdate or tungstate catalyst and an organoaluminum halide cocatalyst. No intermediate resin or cement is produced. Consequently, all reactants and reactant by-products including catalyst metal residues and halide compounds remain in the finished part as contaminants. There is no way to remove these contaminants from the finished article without first destroying it. The metal residues deleteriously affect the electrical insulating properties of the polymer and the halide can combine with moisture to form corrosive hydrogen halide.
To overcome the deficiencies of the ROMP catalyzed thermoplastics it has been proposed to hydrogenate the polymer in an attempt to yield a more stable backbone. However, what is gained in stability is lost in thermal properties. Hydrogenation typically reduces the T.sub.g of the ROMP polymer by approximately 50.degree. C. Furthermore, the cost of the two-step process (ROMP, followed by hydrogenation), the inherent brittleness of the polymer, and the reduced thermal performance of the polymer (T.sub.g &lt;180.degree. C.) is limiting the commercial impact of all ROMP based thermoplastics.
The alternative to the two-step ROMP/hydrogenation route to cyclic olefin polymers is the Ziegler or addition copolymerization route. Addition copolymers derived from higher polycyclic monomers such as tetracyclododecene and ethylene using homogeneous vanadium catalysts are commercially prepared and are available under the trademark Apel.RTM.. However, this catalytic approach can suffer from a number of limitations such as low catalytic activity and significant oligomeric fractions (Benedikt, G. M.; Goodall, B. L.; Marchant, N. S.; Rhodes, L. F. Proceedings of the Worldwide Metallocene Conference (MetCon '94), Catalyst Consultants Inc., Houston, Tex., 1994.)
The limitations of the vanadium catalysts led to the development of the higher activity zirconium-based metallocene polymerization catalysts developed by Prof. Walter Kaminsky (University of Hamburg, Germany) to produce higher molecular weight polycyclic addition copolymers with narrow molecular weight distributions (Plastics News, Feb. 27, 1995, p. 24.). Due to the reduced activity at high polycyclic (norbornene) concentrations, these addition copolymers typically suffer from inadequate T.sub.g s (T.sub.g &lt;240.degree. C.) similar to ROMP catalyzed polymers. Even though these polymers exhibit improved stability, they are still brittle and have poor resistance to hydrocarbon and halohydrocarbon solvents.
Addition homopolymers of norbornene have been polymerized utilizing the Kaminsky zirconium-based metallocene catalysts. These polymers, however, are intractable, e.g., are crystalline, are not soluble in common organic solvents, and do not exhibit a transition (glass or melt) before they decompose (Kaminsky, W.; Bark, A.; Drake, I. Stud. Surf. Sci. Catal. 1990, 56, 425.)
As with the polyimides and BCBs discussed above, the polycycloolefins do not adhere well to metal or silicon surfaces. In order for a polymer to be considered for microelectronic applications, adhesion at different interfaces is a must. A polymer must exhibit satisfactory adhesion to a variety of different substrates, e.g., inorganic substrates such as silicon, silicon dioxide, silicon nitride, alumina, copper, aluminum, and the noble metals such as gold, silver, and platinum, and tie layer metals such as titanium nickel, tantalum, and chromium, as well as to itself when thick layers of the polymer are desired. Good adhesion is required through repeated cycling at temperature extremes (i.e. at depressed and elevated temperatures), as well as through varying humidity conditions. Good adhesion must also be maintained through device processing and assembly temperatures.
Given the adhesion deficiencies inherent with the polyimides and BCBs, techniques have been developed to promote polymer adhesion to surface modified substrates ( i.e., to silicon (oxide) and metal (oxide) substrates). The substrate is typically treated with a difunctional organosilane coupling agent such as .gamma.-amino-propyltriethoxysilane or triethoxyvinylsilane. The polymer or polymer precursor is then cured in contact with the treated substrate. The difunctional organosilane couples the polymer to the substrate. It is believed that the silyl functionality interacts with hydroxyl groups on the substrate surface via hydrolysis to form a covalent linkage as shown below (Soane, D., Martynenko, Z., Polymers in Microelectronics: Fundamentals and Applications, Elsevier, Amsterdam, (1989) 165-169): ##STR5##
The pendant amino functionality (or any suitable functionality) on the treated substrate is then free to react with functional groups on the polymer or polymer precursor to form a bridge that is covalently bonded to the treated substrate and to the polymer. Organosilane treated substrates are disclosed in U.S. Pat. No. 4,831,172 to Hahn et al. and by Heistand II, R. H.; DeVellis, R.; Manial, T. A.; Kennedy, A. P.; Stokich, T. M.; Townsend, P. H.; Garrou, P. E.; Takahashi, T.; Adema, G. M.; Berry, M. J.; and Turlik, I. The International Journal of Microcircuits and Electronic Packaging, 1992, vol. 15, no. 4, 183. See also Polymers In Electronics, supra.
Copper, silver, platinum, and gold are increasingly being utilized in the construction of microelectronic devices as substrates and as conductors due to improved conductivity over the traditionally used aluminum. While organosilane coupling agents are somewhat useful in treating substrates that contain oxygen atoms at the surface, they are noted for poor adhesion to copper, silver, platinum, and especially gold (presumably for the lack of oxygen atoms at the surface interface). Consequently, organosilane coupling agents can only be utilized after a tie-layer has been applied to the substrate surface. Suitable tie-layers include titanium, tantalum, chromium, and nickel. Tie-layers serve to protect the underlying metal substrate, e.g., copper, from polyamic acid (in the case of the polyimides) and/or to provide an adhesion layer which can allow the effective use of the organosilane coupling agents to promote the adhesion of the polyimide. BCBs are not aggressive toward copper so effective organosilane agents can be utilized with copper without a tie-layer. The use of silyl moieties has been shown to be ineffective when gold is employed as the substrate in that amino moieties are required to achieve even modest adhesion at best as demonstrated by Heistand II et al., supra. However, even these typically failed upon exposure at 95.degree. C. in the presence of moisture, suggesting certain failure upon exposure to boiling water.
The use of organosilane treated substrates to promote the adhesion of high performance polymers has its drawbacks. The procedure is a multi-step process requiring additional time and effort to prepare the substrate before the polymer can be applied. First a coupling agent must be applied to the substrate, and then a polymer or polymer precursor with an appropriate co-reactive functional group must be cured or reacted in contact with the surface of the treated substrate. When the use of tie-layers or passivation layers is required, the burden of an extra processing step is added.
There exists a need in the microelectronics industry for a thermally stable, noncorrosive, low dielectric constant polymer with good solvent resistance, high glass transition temperatures, good mechanical performance, and good adhesive properties that can be applied directly to an underlying substrate.
With the inherent low moisture affinity and electrical insulating properties of the addition polymerized polycycloolefins, it would be desirous to improve upon the physical properties (e.g., glass transition temperature, toughness, solvent resistance, etc.) as well as the adhesive properties so that these polymers can be utilized in electrical and optical applications.
The incorporation of functional substituents into hydrocarbon polymer backbones has been a useful method for modifying the chemical and physical properties of the polymer. It is known, however, that polymers containing functional substituents are difficult to prepare because of the propensity of the functional group to poison the catalyst. The free electron pairs on the functional substituent (e.g., nitrogen and oxygen atoms in particular) deactivate the catalyst by complexing with the active catalytic sites. Consequently, catalyst activity decreases and the polymerization of monomer is inhibited.
Previous attempts to addition polymerize a functionally substituted polycycloolefinic monomer via transition metal catalysis have resulted in polymers with low molecular weights. In U.S. Pat. No. 3,330,815 (hereinafter '815), for example, attempts to polymerize functionally substituted polycyclic monomers via palladium metal catalysis produced polymers with low molecular weights as evidenced in the Examples disclosed therein. Molecular weights above 10,000 M.sub.n were not obtained by the disclosed catalyst systems of the '815 patent.
To overcome the difficulty of polymerizing monomers with functional groups (due to catalyst system deactivation), it has been proposed to post react the polymer with the desired functional substituent in order to incorporate the moiety into the polymer. Minami el al., U.S. Pat. No. 5,179,171 (hereinafter '171), disclose copolymers containing ethylenic and polycyclic repeating units which have been post modified with a functional substituent. Among the disclosed copolymers are those derived from maleic anhydride, vinyltriethoxy silane, and glycidyl methacrylate grafted to an ethylenic/polycyclic backbone.
The functional substituent or moiety (i.e. a free radically polymerizable functional group containing monomer) is grafted to the ethylene/polycyclic copolymer by reacting the functional moiety with the base polymer in the presence of a free radical initiator. A free radical moiety (formed from the functional group containing monomer) attacks accessible hydrogens on the polymeric backbone as well as on the polycyclic repeat unit (excluding the bridgehead hydrogens) and grafts to carbon atoms at those sites. The drawback with free radical grafting is that there is no control over where the substituent will graft. Moreover, only small amounts of the free radical moiety (typically less than 2 mole %) grafts to the polymer. Excess amounts of free radical moieties in the reaction medium can cause chain scission, leading to polymer chains of lower molecular weight. There is also a tendency for the grafting moiety to homopolymerize instead of grafting to the base polymer. In addition, grafting monomers have been known to form branched oligomers at the grafting site thereby reducing the efficacy of the desired functionality.
The '171 disclosure purports that the post modification of the disclosed ethylene/polycycloolefin copolymers leads to high T.sub.g polymers (20 to 250.degree. C.). However, the data reported in the Examples appears to suggest otherwise. The maleic anhydride, vinyltriethoxy silane, and glycidyl methacrylate graft copolymers of Examples 33 to 39 on average exhibit a 2.degree. C. increase in T.sub.g over their non-grafted counterparts. When taking experimental error into account, the slight overall increase in the reported T.sub.g values are nil or insignificant at best. Contrary to the disclosure of the '171 patent, high T.sub.g polymers are not attained. In fact, the highest T.sub.g reported in any of the Examples is only 160.degree. C. There is no disclosure in the '171 patent to suggest that addition polymerized silyl substituted polycyclic monomers provide polymers with superior physical and adhesive properties, especially adhesion to copper and noble metal substrates. The data reported in the Examples also indicates the highest incorporation through grafting of the vinyl triethoxy silane moiety to be less than 0.1 mole %.